In the framework of a prospective controlled study in surgical and intensive care units of four study and four control hospitals it was tried to set up an infection control quality management concept for reduction of nosocomial infections in hospital D. With an incidence density of 5.4 nosocomial infections per 1,000 patient days the most favorable situation among the four study hospitals was found initially in this hospital. However, after a small decrease of infection rate during the first study year, no benefit of intervention was achieved in this hospital at the end of the study (6.4 nosocomial infections per 1,000 patient days). Possible explanations may be a very small potential for reduction at the beginning, a change of patient mix due to a new head of the department and the phenomenon of the "regression to the mean". Distinct improvements of the situation were achieved on the level of process quality. Meanwhile surveillance of nosocomial infections became an integral part of quality management in this hospital. The quality circle method was not well accepted for the longterm. But the concept of link nurses was introduced successfully.